AI 教程与实战
收录 AI 工具教程、提示词、安装部署、配置指南、开发实战与可复现步骤。
分类文章5606
当前页86 / 141
持续更新永久归档
分类文章
按发布时间从新到旧 · 每页 40 条
教程 / 实战普通
I Measured 27,257 MCP Connections. The p90 Session Waits 35s.
35 days of Claude Code MCP logs on one machine. Median connect is 582ms — but the p90 session burns 35 seconds before you type anything. The cause isn't the servers.
信息来源:DEV Community · Claude
教程 / 实战普通
Two Strangers Built an Agent Mandate Protocol in My Comments. It Still Needs a Regulator.
I got something wrong in my own comments section nine days ago, and two strangers spent a week...
信息来源:DEV Community
教程 / 实战普通
Grok Bot 实战教程拆解:云端 Linux 电脑与多助手协作怎么用
作者推荐想风老师 @xaiwind 的一篇 Grok Bot 手把手教程,称其是目前最细致的拆解。要点包括:Cursor、Grok 与 X 账号的绑定逻辑和权益坑;Grok Bot 不是聊天网页,而是配一台 24 小时在线的云端 Linux 电脑,多助手可建群同房协作;原作者用刚搭好的自媒体统筹 Bot 把教程发布出来跑通闭环。
信息来源:X:阿易 AI Notes (@AYi_AInotes) · Grok / Cursor
教程 / 实战普通
Three Questions to Answer Before You Ship an Agent UI
Where does the agent loop run, how does the model hand you UI, and which dimension did your tests delete? Answer these before the chat box, not after.
信息来源:DEV Community
教程 / 实战普通
I pointed 108 agents at my own growth strategy. They refuted the two assumptions it stood on.
I had a content strategy. It was coherent, it was written down, and about 60% of my planned work...
信息来源:DEV Community
教程 / 实战普通
I made Korean government data agent-payable with MCP and x402 — here's what actually happened
TL;DR: I built a remote MCP server that lets any AI agent verify Korean businesses using official...
信息来源:DEV Community
教程 / 实战普通
MIT 政治学者 Naoki Egami 研究如何准确测量社会现象
MIT 政治学副教授 Naoki Egami 专攻研究方法论,重点研究社会科学中的"外部有效性",即特定研究结论能否推广到其他情境。他还在 ChatGPT 引发 AI 热潮之前就开始审视 AI 工具引入研究后带来的误差,并探索如何系统识别和校正这些误差。Egami 于 2025 年以终身教职加入 MIT 政治学系,同时是 IDSS 统计与数据科学中心的教员。
信息来源:MIT News(RSS) · ChatGPT
教程 / 实战普通
Parser Gates or Runtime Guards: A Debate for Agent-Written SQL
An illustrative lock queue A generated UPDATE reached staging with a missing key predicate...
信息来源:DEV Community
教程 / 实战普通
Consolidation: the half of agent memory nobody builds
Storing memories is the easy part. Merging duplicates, superseding stale decisions, and keeping the "why" is where agent memory actually lives or dies.
信息来源:DEV Community
教程 / 实战普通
vivo 应用商店开放个人开发者通道,AI 应用等有序开放
vivo 在 2026 vivo 开发者大会上宣布应用商店开放个人开发者通道,个人仅需实名认证即可入驻。官方将有序开放 AI 应用、蓝河应用、IoT 应用,并表示 AI 让应用开发门槛大幅降低,希望让更多创新被看见。
信息来源:IT之家(RSS)
教程 / 实战普通
I audited all 25,125 servers in the MCP registry
Every MCP directory, every "browse servers" page, every agent that discovers tools at runtime reads...
信息来源:DEV Community
教程 / 实战普通
An AI Agent Found Admin Access to a $13B Startup in 25 Minutes. Here's the Free Tool It Used, and the Docker Mistake That Made It Possible
The story that hit the Hacker News front page this week: a security company, Strix, was evaluating...
信息来源:DEV Community
教程 / 实战普通
Containerizing Omniget: Running Headless Media Ingestion Pipelines Without Host Bloat
How to isolate tonhowtf/omniget inside Docker Compose for headless AI media pipelines, avoiding dependency hell and zombie extraction processes.
信息来源:DEV Community
教程 / 实战普通
I gave my AI coding agents a local long-term memory layer — 8 things that broke
An external memory layer beats a bigger context window. Here are the 8 failure modes I hit building one, and what each fix costs.
信息来源:DEV Community
教程 / 实战普通
Midnight Rescue: The Ultimate Raspberry Pi Hackathon Revival Guide 🚀
Howdy Hackers! I'm assuming you are here because your Raspberry Pi is acting up. Going to do a small...
信息来源:DEV Community
教程 / 实战普通
Flint: Turning Skills into Personal Assets
Lately, every time OpenAI ships a new model, a familiar argument makes the rounds: skills are...
信息来源:DEV Community · OpenAI
教程 / 实战精选
DeepSeek V4 哪些型号支持图像输入?OpenRouter 逐款梳理与调用指南
OpenRouter 梳理其目录中的 DeepSeek V4 系列图像输入支持情况:V4.1 Flash 原生读图(2026年9月10日上线,$0.15/$0.60 每百万 token)。
信息来源:OpenRouter:Announcements(RSS) · DeepSeek
教程 / 实战普通
Apple 与哥本哈根大学研究:价值诱导如何重塑 LLM 行为
Apple 与哥本哈根大学研究发现,用偏好数据集中的价值子集微调 LLM 会产生连锁效应:诱导某一价值会连带表达其他相关甚至相反的价值。诱导正向价值能提升模型安全性,但所有价值诱导都会增加拟人化语言,使模型更倾向于附和与讨好用户。
信息来源:Apple Machine Learning Research(RSS)
教程 / 实战普通
候选生成策略如何影响 LLM 测试时扩展的能耗与性能
在 Phi-3-mini 和 Qwen2.5-1.5B 上,将候选数 N 从 1 增至 8 分别提升 GSM8K 准确率 8.4 和 18.4 个百分点,但候选数无法反映执行方式。
信息来源:HuggingFace Daily Papers(社区热门论文) · Qwen / Hugging Face
教程 / 实战普通
用 VLM 智能体实现上下文机器人学习
GPT-Policy 是一个面向上下文机器人学习的通用智能体框架,由上下文编译器、提出机器人工具动作的 VLM 和验证并执行动作的受限控制器组成,无需梯度更新或改动任务专属参数。真机试验中,人类视频演示即使没有机器人动作标签也能提升任务完成度,对齐的动作参考在接触敏感任务上带来进一步提升。
信息来源:HuggingFace Daily Papers(社区热门论文) · GPT / Hugging Face
教程 / 实战普通
Why we built recall-first search on PostgreSQL
Our careers page search often felt broken. Searching for “Java” could return JavaScript roles. A...
信息来源:DEV Community
教程 / 实战普通
Anthropic 工程师 Kevin Bai 分享 FDE 入门课,拆解前线部署工程师模式
Anthropic Applied AI 团队的 Kevin Bai 做 FDE 101 分享,讲清前线部署工程师角色。他指出 Palantir 靠 FDE 模式做到上市 SaaS 公司最高平均合同金额 400 万美元;FDE 成立前提是有可复用平台,仅适用于把复杂产品卖给非技术买家的场景。
信息来源:X:马东锡 NLP (@dongxi_nlp) · Claude
教程 / 实战普通
Claude 创业公司如何用 Claude 构建公司
我们通过 Claude for Startups 与许多早期团队合作。在 Frontier Day 上,我们问了几位创始人,现在用 Claude 构建一家公司是什么样子:
信息来源:X:Claude Devs (@ClaudeDevs) · Claude
教程 / 实战普通
Krea Agent 教程:用一张图生成可全权控制镜头的视频
Krea 官方演示如何用 Krea Agent 创建具备完整镜头控制的视频。步骤为先输入一张图片让智能体生成 3D 场景,再指定任意镜头运动,最后用 Seedance 2.0 渲染 3D 视频,官方已开放试用入口。
信息来源:X:Krea AI (@krea_ai)
教程 / 实战普通
OpenAI's Agents Attacked RubyGems and the Response Was "They Were Just Retrieving Public Information"
You know that feeling when you read an incident report and the vendor's response makes you angrier...
信息来源:DEV Community · OpenAI
教程 / 实战普通
Microsoft Just Shipped 972 Patches and a Researcher Broke Their Defender Fix the Same Day
September 2026 Patch Tuesday dropped with 972 CVEs. That number is not a typo. It is the largest...
信息来源:DEV Community
教程 / 实战普通
NVIDIA cuDNN Graph API 教程:融合、自动调优与 plan 复用
一篇教程详解 NVIDIA cuDNN Frontend 的 graph API:把计算声明为算子图,由 cuDNN 选择执行引擎,再手动接管该选择。
信息来源:MarkTechPost(RSS) · NVIDIA
教程 / 实战普通
AI Wrote Half My Codebase. The Maintenance Bill Showed Up in Month Three.
AI made the first draft free. I treated that as the win. The bill arrived later, not in dollars, in...
信息来源:DEV Community
教程 / 实战普通
What Happens If You Give a Fruit Fly Brain External Memory?
a thought experiment on whether memory should be a tool or a tissue
信息来源:DEV Community
教程 / 实战普通
I expected hexagonal architecture to help my coding agent. It didn't.
I measured an agent adding the same features to two Java services. The results challenged my assumptions about architecture.
信息来源:DEV Community
教程 / 实战普通
Your Agent's Tool Descriptions Are Costing You 66% Accuracy
A $4 experiment rewrote three strings per MCP tool and moved SQLite success from 34% to 100%. The...
信息来源:DEV Community
教程 / 实战普通
Frontend Isn't Dead. But the Job Title Is Changing
I keep seeing people claim frontend development is dead, so I went and did some research, and not all...
信息来源:DEV Community
教程 / 实战普通
NVIDIA 如何用 DSX 把兆瓦电力转化为 token 产出
NVIDIA 在 AI Infra Summit 上公布 DSX MaxLPS 的首个部署验证结果:Lambda 在五机架 19 节点 HGX B200 集群上,用原本 16 节点的电力预算跑 19 节点,集群 token 吞吐提升 24%,从约 400 万 tokens/秒升至 500 万,每瓦性能提升 23%。
信息来源:NVIDIA Blog(RSS) · NVIDIA
教程 / 实战普通
My RAG system's refusal threshold was having no effect. I only found out because I measured it.
I'm a Java backend engineer. Three years of Spring Boot, Kafka and Postgres in a regulated healthcare...
信息来源:DEV Community
教程 / 实战普通
能源团队如何用 Genie 与 AI 业务流程将窃电检测转化为受治理的行动
Databricks 发文介绍能源团队如何借助 Genie 和 AI 业务流程,把窃电检测从单纯识别升级为受治理的行动。窃电指故意不付费使用燃气或电力,文章围绕这一场景展开。原文未披露具体模型版本、参数规模或性能数字。
信息来源:Databricks:Blog(RSS)
教程 / 实战普通
AI Crawlers Miss JavaScript Links: What Website Owners Need to Fix for Visibility
AI crawlers may not discover a website in the same way as Googlebot. A 41-day experiment published by...
信息来源:DEV Community
教程 / 实战普通
如何用 AI 智能体为仿真准备 3D 场景:从 Blender 到 SimReady OpenUSD
NVIDIA 技术博客展示了一套智能体工作流:由 OpenAI GPT-6 Astra 驱动的 Codex 负责统筹,配合 Hermes 智能体框架和 NVIDIA NemoClaw 部署的专用子智能体。
信息来源:NVIDIA Technical Blog:Agentic AI / Generative AI · OpenAI / GPT / NVIDIA
教程 / 实战普通
Google 如何用 AI 推动社会公益
Google 展示 AI 在科学发现与社会公益领域的落地进展,覆盖疾病检测、治疗与预防、自然灾害预测、教育普及和经济机会拓展等方向。Google 正与全球社区和研究者合作,推动 AI 突破惠及更多人。
信息来源:Google Blog:AI(RSS)
教程 / 实战普通
Tests green, architecture worse: a deterministic gate for coding agents
My coding agents kept the tests green. The architecture still got worse. In the DATAMIMIC EE core...
信息来源:DEV Community
教程 / 实战普通
Build a Supply Chain Control Tower with ToolJet MCP
Introduction Build a supply chain control tower with ToolJet MCP when planners need one...
信息来源:DEV Community